• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Phase Distortion ABX testing

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
1,725
Likes
865
Location
Salvador-Bahia-Brasil
I still don't find the frequency dependent windowed SPL magnitude graphs all that informative. It's far better to look at the wavelet transform if you want to see the obvious improvement in both the time and frequency domains at the same time.

FDW windowing is not about looking at the time domain. it is about trying to get a psychoachoustic FR graph. the normal window doesn't show what we hear since above 1000Hz-ish our ears will seperate direct sound from later-ish reflections.
this is how FDW entered REW: https://www.hometheatershack.com/threads/feature-request-frequency-dependent-windowing.99673/

if there's anything that may indicate "bad pre-ringing" in the measurements.

https://www.audiosciencereview.com/...via-rephase-and-camilladsp.19910/#post-656041
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
1,634
Likes
1,140
Location
Canada
maybe you can explain when you play the guitar test original on your speakers. so you need not recrord it. which sound in stereo width between the KH120 and sceptre s8 is larger or are they same ?. or which speaker sound more simular crisp to headphone. https://www.audiosciencereview.com/...tep-response-and-audio-record-examples.19812/ here you can download original. Recording of a speaker is usefull because you can then do blind tests with the free version
https://hofa-plugins.de/en/plugins/4u-blindtest/non headphones. hear the audio examples of kali and small woofer speakers. the diffrence is very large.better phase i hear give a little better room feeling on kali but it is not so much as need. test with distort guitar shows best slow speakers because when see wave of recordet guitar the levels change much (but not on high frequency). so can hear good when a speaker is too slow. maybe you hear this not as reduce stereo width but as muddy sound(when both speakers are frequency correct)

When comparing between two different speakers, you aren’t hearing only isolated effects from their transient response as indicated by the step... so using the step response and/or impulse view as an indicator of stereo-width or soundstage quality is really stretching things. Remember, the step, impulse, and ETC views have little frequency domain information! You cannot determine which speaker overall will have better performance in rendering stereo soundstage by simply looking at its step response — esp. if examining only one direction or plane.

When it comes to listening with whatever gear and setup conditions one has on hand, subjectivity also reins supreme... we already know some people barely can hear any change at all that’s worth fussing about. Personally, I don’t really care if anyone passes a blind test here or not — I was already happy with my analog KH120 speakers even way before I “linearized” them. They only sounded dull when directly listening side-by-side with the S8 in one particular listening session with certain particular tracks — not something I care enough to lose any more sleep over.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
1,634
Likes
1,140
Location
Canada
FDW windowing is not about looking at the time domain. it is about trying to get a psychoachoustic FR graph. the normal window doesn't show what we hear since above 1000Hz-ish our ears will seperate direct sound from later-ish reflections.
this is how FDW entered REW: https://www.hometheatershack.com/threads/feature-request-frequency-dependent-windowing.99673/



https://www.audiosciencereview.com/...via-rephase-and-camilladsp.19910/#post-656041

Fascinating thread. Will finish reading and mulling over it later. As you probably already know, my understanding of the applications and origins of FDW is limited. I wasn’t under the impression that it was developed simply as a means to generate better “psychoacoustic” frequency magnitude graphs — though, certainly that’s part of it. I had other uses mainly in mind for this method of filtering such as reducing visible effects of reflections and late arriving energy in the total recorded response for easier reading or analysis of graphs — weeding out extraneous “noise” — which may just happen to be “psychoacoustically” less important, if you will.
 

bennybbbx

Senior Member
Joined
May 27, 2020
Messages
368
Likes
78
Location
germany
When comparing between two different speakers, you aren’t hearing only isolated effects from their transient response as indicated by the step... so using the step response and/or impulse view as an indicator of stereo-width or soundstage quality is really stretching things. Remember, the step, impulse, and ETC views have little frequency domain information! You cannot determine which speaker overall will have better performance in rendering stereo soundstage by simply looking at its step response — esp. if examining only one direction or plane.

When it comes to listening with whatever gear and setup conditions one has on hand, subjectivity also reins supreme... we already know some people barely can hear any change at all that’s worth fussing about. Personally, I don’t really care if anyone passes a blind test here or not — I was already happy with my analog KH120 speakers even way before I “linearized” them. They only sounded dull when directly listening side-by-side with the S8 in one particular listening session with certain particular tracks — not something I care enough to lose any more sleep over.

at least step response show unprecise speakers. seem a reason wy speakers sound worse in stereo width. here is something about binaural heraring. remember 22 microseconds are 44 khz. maybe i should do blind delay tests how good users in this forum hear.
Abstract

The auditory system encodes the timing of peaks in basilar-membrane motion with exquisite precision, and perceptual models of binaural processing indicate that the limit of temporal resolution in humans is as little as 10-20 microseconds. In these binaural studies, pairs of continuous sounds with microsecond differences are presented simultaneously, one sound to each ear. In this paper, a monaural masking experiment is described in which pairs of continuous sounds with microsecond time differences were combined and presented to both ears. The stimuli were matched in terms of the excitation patterns they produced, and a perceptual model of monaural processing indicates that the limit of temporal resolution in this case is similar to that in the binaural system.

https://pubmed.ncbi.nlm.nih.gov/12765396/
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
1,725
Likes
865
Location
Salvador-Bahia-Brasil
ok, here are the 3, var smoothing, aligned at 1000Hz

View attachment 109871

and with 5 cycle FDW

View attachment 109872

5 cycle FDW of the the two corected only

View attachment 109873




OBS: no SPL calibration
for the info: the goal of this filters was flat direct sound

ok, after reading the hometheatershack topic it suddenly hit me. If you limit the window all delayed frequencies will be distorted.
I took minimum phase exports of both the measurements, reimported them, and now the actualy look the same (5 cycles again):

frefer.jpg


so when my manual Rephase phase correction on the minimum phase corrected version showed a diferent FR in FDW, it didn't realy change the FR, but brought more of the response into the window
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
1,725
Likes
865
Location
Salvador-Bahia-Brasil
Fascinating thread. Will finish reading and mulling over it later. As you probably already know, my understanding of the applications and origins of FDW is limited. I wasn’t under the impression that it was developed simply as a means to generate better “psychoacoustic” frequency magnitude graphs — though, certainly that’s part of it. I had other uses mainly in mind for this method of filtering such as reducing visible effects of reflections and late arriving energy in the total recorded response for easier reading or analysis of graphs — weeding out extraneous “noise” — which may just happen to be “psychoacoustically” less important, if you will.

it's funny that Bob Katz actualy gave up on the topic. I tried revitalizing it with my "finding" above
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
1,725
Likes
865
Location
Salvador-Bahia-Brasil
also, the more I think of it, the less sense makes the fixed window (500ms). think about it, a room with a lot of ringing in the bass will add a lot of this later response to the overall FR. this might be the reason why some prefere the Harman bass boost. It might just be the noise in the FR which is spread out in the time domain. A heavy treated room will add only a small part of this to the overall FR
 
OP
B

BenB

Active Member
Joined
Apr 18, 2020
Messages
185
Likes
269
ok, after reading the hometheatershack topic it suddenly hit me. If you limit the window all delayed frequencies will be distorted.
I took minimum phase exports of both the measurements, reimported them, and now the actualy look the same (5 cycles again):



so when my manual Rephase phase correction on the minimum phase corrected version showed a diferent FR in FDW, it didn't realy change the FR, but brought more of the response into the window

I was going to make that argument. It was nothing more than a measurement artifact based on the impact of arbitrary parameters being applied to the processing. I'm glad you got it corrected, and posted about. I'm not convinced that ANY of this actually belongs in a thread about ABX phase distortion testing, though.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
1,725
Likes
865
Location
Salvador-Bahia-Brasil
I'm not convinced that ANY of this actually belongs in a thread about ABX phase distortion testing, though.

not anymore, but at the time of posting I was convinced it was ontopic since I interpreted the FR diferences as phase distorsion
 

vair

Member
Joined
Nov 10, 2020
Messages
6
Likes
1
Wooww interesting post ! Thank you :)

I've been searching in this area for some months, trying to understand the audibility of group delay "issues".

Regarding the ABX test, I have to say that I could not hear anything on headphones, and it was quite demanding to hear it on speakers (apollo twin, macintosh amp, atohm eurus, partially treated room)..

I managed to distinguish them with relatively high confidence using this technique : play the "dry" file first (2 times), then the X.
It has worked great this way, I can distinguish something slightly different in the "impact sensation". The dry one seems more "tight", with a tiny bit more punch (chest sensation).

Without this "trick", I found mysef unable to distinguish the two.. Listening to dry then all-passed one alternatively I completely loose my marks and results were purely chance.. interesting isn't it ?

Also, it strucked me that I was way more "coherent" without thinking too much "intellectually". I had better results closing my eyes, emptying my brain and answering intuitively.. that was a strange sensation..

Finally, I was unable to hear it after too long listening session.. but it was ok after a short pause.

Might post some results if you find it interesting, and will do another session in the day to confirm/infirm these first results.
 
Last edited:

vair

Member
Joined
Nov 10, 2020
Messages
6
Likes
1
Did a rough analysis of the kick drum sample files. As far as I can tell, nothing conspicuous can be detected.

...

Next, let's look at the phase frequency response of the kick drum samples. Around 100Hz, the difference between the unmodified sample and the one with the 4th order allpass should be 180°.

Unmodified sample (with unwrap phase - see green dots):
View attachment 107734

Sample with allpass 4th order (with unwrap phase - see blue dots):
View attachment 107735

The phase shift is almost exactly 180°. So also here everything is as described by @BenB

Except for the slight sound pressure deviation, I can't detect any error in the kick drum samples.
The question now is whether you can hear the small sound pressure deviation in the ABX test, or whether the phase shift is audible?
.

I was curious to understand the meaning of phase measurement on an audio file. How can we give a phase value without refering to another signal ?
Then the start of the audio file is taken as a time cue ? Would it be the same in a Fourrier analysis ?





Also, comparing the kicks spectrogram is interesting (the one that moves right is the allpassed)

spectrogram.gif


Maybe a reason why it is so hard to distinguish may be that the "overall shape" isn't affected that much by the all pass ?


Looking at the impulse is also interesting (light red is allpassed) :

kick impulse.gif


Maybe a reason why it is indeed possible to distinguish the two on speaker is that the first transients seems slightly affected, with less energy ?



Anyway we're speaking about subtle things here..
 
Last edited:

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
1,725
Likes
865
Location
Salvador-Bahia-Brasil
I can distinguish something slightly different in the "impact sensation". The dry one seems more "tight", with a tiny bit more punch (chest sensation).

this is the main diference with a linear group delay. the effect is manly on the transients. it makes sense if you think about it. having part of the transients delayed will have an effect like a compressor with short atack.
with that beeing said, how big the impact is depends on the music style
 
Top Bottom